摘要
将视觉原理和生物物理学中著名的Weber定律结合在一起,提出一个基于视觉原理的新算法,实现了有效而且无参型聚类,并且提出了相应的新代价标准.仿真实验表明对于非线性—不可分的数据集,该新聚类算法有效,获得了传统聚类算法(如FCM)所达不到的效果.
The main contribution of this paper is to present a new visual theoretic clustering algorithm, which integrates visual systems together with the famous Weber law in biophysics to realize effective and nonparametric clustering. A new cost criterion for clustering is also presented. Our simulations demonstrate that this new nonparametric clustering algorithm is effective for nonlinearly separable datasets which in general the conventional clustering approaches such as FCM can not well cope with.
出处
《江南大学学报（自然科学版）》
CAS
2004年第1期9-14,共6页
Joural of Jiangnan University (Natural Science Edition)
基金
江苏省自然科学基金项目(BK2003017)资助课题.
